Software is often thought of like a manufacturing industry. Companies make a ‘product’ and sell many copies. Partly because this is the model that IP induces (IP being a structure of old material manufacturing industry).
But there seem to be simple reasons why this does not fit. Software is different because the basic economic ingredients are different. And this points toward a rather different ideal economic arrangement.
Software is based on design (all kinds: engineering, UI, etc.), not manufacturing.
For manufacturing industry, substantial cost is in the manufacturing – you have to build each item, and the more you produce the bigger factory you need.
But for software, the cost is almost entirely in the design. And design is not construction- or material-bound, but information-bound
Manufacturing has economy of scale, but design has more like the opposite: diseconomy of scale.
You can only design with/for information you have in front of you. It is difficult to get information from everywhere and everyone on what they want, or information on all possible uses, or of what the future might require.
On the other hand, distribution of the designed software scales almost ideally. Copies have negligible cost, but duplicate all their value – each gives its entire value to the user.
Software is about providing good services, not producing copies of goods.
It follows from the above that the best arrangement is: 1, design only with information readily accessible; 2, make copies as easily/widely/freely available as possible.
Software work should be on focused projects: significantly ‘localised’ or limited. Not necessarily for small groups of customers, but for clearly defined aims.
One might say that the customer's payment should go to the ‘marginal cost of design’ – the design effort needed to take from currently freely available parts and reuse and make them into something particular and specific.
This removes risk because it is a good use of information. It covers the cost of design in a complete and sure way, and allows the unpredictable value of the software to be completely realised. In contrast, mass manufacturing a heap of items is a cost based on uncertain knowledge, and selling copies is a restriction of unexpected value
It cannot be too far wrong because it reflects some of what is actually happening.
One could perhaps say that all software development does to some extent follow this form. An average example might be web-development: customers (from individuals to large companies) buy particular web site development – custom software design work based on reusing/remaking freely available parts.
What is lacking is a well-developed way to connect sellers and buyers in collective way. People have shared demands, and ought to be able to share the cost of appropriate production.